Friday, September 23, 2011

Mirror Neuron Forum - Role of mirror neurons in speech and language processing - Part I

Now on to my favorite mirror neuron topic, Question 2 of the Mirror Neuron Forum:

Do Mirror Mechanisms Causally Contribute to Speech Perception and Language Comprehension?

There are two questions here, each logically independent of the other, but findings from one domain may provides hints regarding the other. The first is whether mirror neurons are the basis of speech sound recognition. This was the first language-related ability that mirror neuron function was generalized to in humans. The second question is whether the motor system -- often defined as the somatotopically organized fields such as M1, which is generally consider NOT to be part of the mirror system, but no one seems to worry about that for some reason -- is involved in the representation of action-related concepts. One question is a perceptual issue, the other is a semantic/conceptual issue.

I focused on the first question for two reasons. One is its primacy in the history of the development of theories of mirror neuron function in humans. The second is that there is a TON of data on the topic, allowing us to draw firm conclusions. I consider this a test case for the MN theory and suggested that if the theory fails here, we need to seriously question its role in other domains. I then presented a list of the evidence proving (I almost never use this word, but I think it is justified here) that the motor speech system is NOT necessary for speech recognition.

Gallese did not dispute this claim. Instead he questioned whether findings from the speech perception literature should lead us to question findings in other domains.

VG: According to GH, the roles of MNs in speech perception and language understanding are to be considered tightly related: If a relationship between MNs and speech perception cannot be established, so the argument goes, it would follow that the connection between MNs and language understanding would be falsified. I disagree with this logic.

Note that I didn't actually say that findings from speech perception would falsify claims regarding language semantics. I said, "If the action understanding interpretation fails for speech perception, it raises serious questions about the theory generally." Why do I say this? Because this is the domain in which we have the most evidence. It is a test case. If the theory holds up for speech perception, then it passed a rigorous test and we might be more lenient in accepting weaker data in other domains. If it fails the rigorous test, this leads us to question the weaker data. Could data from other studies lead to the firm conclusion that motor systems play a role in action knowledge representation (or empathy or whatever)? Yes. But we are not there yet. Speech perception is the ONLY domain where the results are conclusive and the theory failed. This was my point.

Now, in another recently published paper, I reviewed the evidence claimed to support the theory that the motor system is critically involved in action semantics and found the evidence weak at best (Hickok, 2010). So let's look at what Gallese takes to be some of the strongest claims.
VG: In humans, the cortical motor system is activated during the observation of a variety of motor behaviors
Activation does not imply causation.
VG: right handers preferentially activate the left premotor cortex during lexical decisions on manual-action verbs (compared with nonmanual-action verbs), whereas left handers preferentially activate right premotor areas (Willems, Hagoort,&Casasanto, 2010). Thus, right and left handers, who performactions differently, use correspondingly different areas of the brain for representing action verb meanings
That's nice but unsurprising and easily explainable without assuming that the meaning of the verbs is coded in the motor system. If I say a word like throw this will activate in your brain a network of systems and representations that have previously been associated with that word. Chances are, you have previously linked that word with the very action itself: "Throw me the ball!" upon which you generate the movement. So even if the movement itself is not part of the meaning of the word, motor programs for generating the movement just might activate when you hear the word. So given that lefties and righties throw with different hands, you would expect to see the observed difference. Depending on your recent life experiences, upon hearing throw you might also activate the word up and the memory of a wild party, but that doesn't mean that up and WILD PARTY are part of the meaning of to throw, it just means they are associated at some level.

How can we test this idea more directly? One prediction is that damage to the motor system should cause deficits in understanding actions. Some studies have been published which are suggestive in this direction, e.g., in Parkinson's patients, but these cases are far from complication free as I noted in my 2010 review. Unfortunately, there is not a lot of (convincing) experimental evidence available. However, I will again point out that we can readily understand actions that we cannot perform such as the coiling of a snake or the flying of a bird. Further, from an evolutionary standpoint, these are actions that are critical to understand because survival can depend on it. This indicates that action understanding, at a fundamental level, cannot be dependent on motor representations. So to sum up: the MN theory of action understanding has failed its only rigorous test. The evidence supporting the role of MNs in action semantics is debatable. There is evidence that the motor system is not critical for understanding actions generally. Together, this leads me to "seriously question" the claim that actions semantics depends on the motor system.

Gallese, V., Gernsbacher, M., Heyes, C., Hickok, G., & Iacoboni, M. (2011). Mirror Neuron Forum Perspectives on Psychological Science, 6 (4), 369-407 DOI: 10.1177/1745691611413392

Hickok, G. (2010). The role of mirror neurons in speech perception and action word semantics. Language and Cognitive Processes, 25, 749 - 776.

No comments: